Changed models to use glossarist v2 format

lib/termium.rb

@@ -3,10 +3,10 @@
 require "glossarist"
 
 require "lutaml/model"
-require "lutaml/model/xml_adapter/nokogiri_adapter"
+require "lutaml/model/xml/nokogiri_adapter"
 
 Lutaml::Model::Config.configure do |config|
-  config.xml_adapter = Lutaml::Model::XmlAdapter::NokogiriAdapter
+  config.xml_adapter = Lutaml::Model::Xml::NokogiriAdapter
 end
 
 module Termium

lib/termium/abbreviation.rb

@@ -52,5 +52,17 @@ def to_h
 
       set
     end
+
+    def to_designation
+      attrs = {
+        designation: value,
+        normative_status: deprecated ? "deprecated" : "preferred",
+      }
+
+      attrs[:gender] = gender if gender
+      attrs[:part_of_speech] = part_of_speech if part_of_speech
+
+      Glossarist::Designation::Abbreviation.new(attrs)
+    end
   end
 end

lib/termium/core.rb

@@ -47,10 +47,13 @@ def uuid(str = identification_number)
     # details="Compartment - ISO/IEC JTC 1 Information Technology Vocabulary" />
     def to_concept(options = {})
       Glossarist::ManagedConcept.new.tap do |concept|
-        # The way to set the universal concept's identifier: data.identifier
-        concept.id = identification_number
+        # V2: Create new data object to ensure it's serialized (not marked as default)
+        concept.data = Glossarist::ManagedConceptData.new(
+          id: identification_number,
+          sources: concept_sources
+        )
 
-        concept.uuid = uuid
+        concept.id = uuid
 
         # Assume no related concepts
         concept.related = []
@@ -60,19 +63,19 @@ def to_concept(options = {})
           concept.date_accepted = options[:date_accepted]
         end
 
-        language_module.map do |lang_mod|
+        language_module.each do |lang_mod|
           localized_concept = lang_mod.to_concept(options)
 
           # TODO: This is needed to skip the empty french entries of 10031781 and 10031778
           next if localized_concept.nil?
 
-          localized_concept.id = identification_number
-          localized_concept.uuid = uuid("#{identification_number}-#{lang_mod.language}")
+          localized_concept.data.id = identification_number
+          localized_concept.id = uuid("#{identification_number}-#{lang_mod.language}")
 
           universal_entry.each do |entry|
-            localized_concept.notes << Glossarist::DetailedDefinition.new(content: entry.value)
+            localized_concept.data.notes << Glossarist::DetailedDefinition.new(content: entry.value)
           end
-          localized_concept.sources = concept_sources
+          localized_concept.data.sources = concept_sources
           concept.add_localization(localized_concept)
         end
       end

lib/termium/entry_term.rb

@@ -86,5 +86,19 @@ def to_h
 
       set
     end
+
+    def to_designation
+      attrs = {
+        designation: value,
+        normative_status: normative_status,
+      }
+
+      attrs[:geographical_area] = geographical_area if geographical_area
+      attrs[:plurality] = plurality if plurality
+      attrs[:gender] = gender if gender
+      attrs[:part_of_speech] = part_of_speech if part_of_speech
+
+      Glossarist::Designation::Expression.new(attrs)
+    end
   end
 end

lib/termium/extract.rb

@@ -30,8 +30,9 @@ class Extract < Lutaml::Model::Serializable
 
     def to_concept(options = {})
       coll = Glossarist::ManagedConceptCollection.new
-      coll.managed_concepts = core.map do |managed_concept|
-        managed_concept.to_concept(options)
+      core.each do |managed_concept|
+        concept = managed_concept.to_concept(options)
+        coll.store(concept)
       end
       coll
     end

lib/termium/language_module.rb

@@ -47,7 +47,7 @@ def abbreviations
 
     def designations
       # NOTE: entry_term is a collection
-      entry_term + abbreviations
+      (entry_term + abbreviations).compact
     end
 
     def to_h
@@ -69,17 +69,22 @@ def to_h
     end
 
     def to_concept(options = {})
-      x = to_h
-      return nil unless x
+      return nil unless definition
+
+      Glossarist::LocalizedConcept.new.tap do |concept|
+        concept.data = Glossarist::ConceptData.new(
+          language_code: LANGUAGE_CODE_MAPPING[language.downcase],
+          terms: designations.map(&:to_designation),
+          definition: [Glossarist::DetailedDefinition.new(content: definition)],
+          notes: notes.map { |n| Glossarist::DetailedDefinition.new(content: n) },
+          examples: examples.map { |e| Glossarist::DetailedDefinition.new(content: e) },
+          entry_status: "valid",
+          domain: domain
+        )
 
-      Glossarist::LocalizedConcept.new(x).tap do |concept|
-        # Fill in register parameters
         if options[:date_accepted]
-          puts options[:date_accepted].inspect
           concept.date_accepted = options[:date_accepted]
         end
-
-        puts concept.inspect
       end
     end
   end

lib/termium/source.rb

@@ -25,11 +25,11 @@ def content
     end
 
     def to_concept_source
-      Glossarist::ConceptSource.new({
-                                      "type" => "lineage",
-                                      "ref" => content,
-                                      "status" => "identical",
-                                    })
+      Glossarist::ConceptSource.new(
+        type: "lineage",
+        status: "identical",
+        origin: Glossarist::Citation.new(ref: content)
+      )
     end
   end
 end

spec/fixtures/expected_v2_output/concept/4d2d4ac4-af47-545c-9d41-19285d785fc5.yaml

@@ -0,0 +1,17 @@
+---
+data:
+  identifier: '2123225'
+  localized_concepts:
+    eng: 33f0da1d-b8af-511c-8a7a-2f777419fa95
+    fre: 848965b7-2090-5f93-af30-8cba405c248c
+  sources:
+  - origin:
+      ref: ISO/IEC 2382-16:1996
+    status: identical
+    type: lineage
+  - origin:
+      ref: Ranger, Natalie * 2006 * Bureau de la traduction
+    status: identical
+    type: lineage
+id: 4d2d4ac4-af47-545c-9d41-19285d785fc5
+status: valid

spec/fixtures/expected_v2_output/localized_concept/33f0da1d-b8af-511c-8a7a-2f777419fa95.yaml

@@ -0,0 +1,25 @@
+---
+data:
+  definition:
+  - content: quotient of the character mean entropy by the mean duration of a character
+  examples: []
+  id: '2123225'
+  notes:
+  - content: The average information rate may be expressed in shannons per second.
+  - content: 16.04.07 (2382)
+  sources:
+  - origin:
+      ref: ISO/IEC 2382-16:1996
+    status: identical
+    type: lineage
+  - origin:
+      ref: Ranger, Natalie * 2006 * Bureau de la traduction
+    status: identical
+    type: lineage
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: average information rate
+  language_code: eng
+  entry_status: valid
+id: 33f0da1d-b8af-511c-8a7a-2f777419fa95

spec/fixtures/expected_v2_output/localized_concept/848965b7-2090-5f93-af30-8cba405c248c.yaml

@@ -0,0 +1,26 @@
+---
+data:
+  definition:
+  - content: quotient de l'entropie moyenne par caractère par la durée moyenne d'un
+      caractère
+  examples: []
+  id: '2123225'
+  notes:
+  - content: Le débit moyen d'entropie peut s'exprimer en shannons par seconde.
+  - content: 16.04.07 (2382)
+  sources:
+  - origin:
+      ref: ISO/IEC 2382-16:1996
+    status: identical
+    type: lineage
+  - origin:
+      ref: Ranger, Natalie * 2006 * Bureau de la traduction
+    status: identical
+    type: lineage
+  terms:
+  - type: expression
+    normative_status: preferred
+    designation: débit moyen d'entropie
+  language_code: fre
+  entry_status: valid
+id: 848965b7-2090-5f93-af30-8cba405c248c

spec/fixtures/single_entry.xml

@@ -0,0 +1,45 @@
+<?xml version='1.0' encoding='UTF-8' standalone='yes'?>
+<ns2:termium_extract xmlns:ns2="http://termium.tpsgc-pwgsc.gc.ca/schemas/2012/06/Termium" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" language="EN" xsi:schemaLocation="http://termium.tpsgc-pwgsc.gc.ca/schemas/2012/06/Termium http://termium.tpsgc-pwgsc.gc.ca/schemas/2012/06/Termium.xsd">
+<extractLanguage language="EN" order="0"/>
+<extractLanguage language="FR" order="1"/>
+<core identificationNumber="2123225" disseminationLevel="P">
+  <languageModule language="EN">
+    <entryTerm order="1" value="average information rate">
+      <sourceRef order="1"/>
+      <parameter abbreviation="COR"/>
+      <parameter abbreviation="NORM"/>
+    </entryTerm>
+    <textualSupport order="1" type="DEF">
+      <value>quotient of the character mean entropy by the mean duration of a character</value>
+      <sourceRef order="1"/>
+    </textualSupport>
+    <textualSupport order="2" type="OBS">
+      <value>The average information rate may be expressed in shannons per second.</value>
+      <sourceRef order="1"/>
+    </textualSupport>
+  </languageModule>
+  <languageModule language="FR">
+    <entryTerm order="1" value="débit moyen d'entropie">
+      <sourceRef order="1"/>
+      <parameter abbreviation="COR"/>
+      <parameter abbreviation="M"/>
+      <parameter abbreviation="NORM"/>
+    </entryTerm>
+    <textualSupport order="1" type="DEF">
+      <value>quotient de l'entropie moyenne par caractère par la durée moyenne d'un caractère</value>
+      <sourceRef order="1"/>
+    </textualSupport>
+    <textualSupport order="2" type="OBS">
+      <value>Le débit moyen d'entropie peut s'exprimer en shannons par seconde.</value>
+      <sourceRef order="1"/>
+    </textualSupport>
+  </languageModule>
+  <universalEntry order="1">
+    <value>16.04.07 (2382)</value>
+    <sourceRef order="1"/>
+    <parameter abbreviation="CONUM"/>
+  </universalEntry>
+  <source order="1" details="ISO-IEC-2382-16 * 1996 *  *  * "/>
+  <source order="2" details="Ranger, Natalie * 2006 * Bureau de la traduction"/>
+</core>
+</ns2:termium_extract>